We investigated whether random proinsulin levels and proinsulin:C-peptide ratio (PI:C) complement immune and genetic markers for identifying relatives at high risk of type 1 diabetes. During an initial sampling, random glycaemia, proinsulin, PI:C and HLA DQ genotype were determined in 561 non-diabetic first-degree relatives who had been positive for islet autoantibodies on one or more occasions and in 561 age- and sex-matched persistently antibody-negative relatives. During follow-up (median 62 months), 46 relatives with antibodies at entry developed type 1 diabetes. At baseline, antibody-positive relatives (n=338) had higher PI:C values (p<0.001) than antibody-negative subjects with (n=223) or subjects without (n=561) later seroconversion. Proinsulin and PI:C were graded according to risk of diabetes as expressed by positivity for (multiple) antibodies or IA-2 antibodies, especially in persons carrying the high-risk HLA DQ2/DQ8 genotype and in prediabetic relatives. In the presence of multiple or IA-2 antibodies, a PI:C ratio exceeding percentile 66 of all antibody-negative relatives at entry (n=784) conferred a 5-year diabetes risk of 50% and 68%, respectively (p<0.001 vs 13% for same antibody status with PI:C<percentile 66). Cox regression analysis confirmed random PI:C as an independent predictor of the risk of diabetes (p< or =0.001). Random proinsulin and PI:C represent dynamic markers of the state of beta cell function that complement immune markers in identifying relatives who are at homogeneously high risk of contracting type 1 diabetes and are therefore eligible for secondary prevention trials.